Summary of findings
This study estimated complex inequalities and horizontal inequities in primary and secondary health care utilization in Northern Sweden at the intersection of gender and education. The results illustrate the complexity and unique evidence arising from applying an intersectional perspective.
First, we did not find robust evidence for any excess disparity of double (dis)advantage, but rather that the axes of gender and education were independently expressed in health care utilization disparities. Second, low-educated women utilized primary and secondary care considerably more frequently than men, but this inequality was largely (primary care) or completely (secondary care) explained by the greater health care needs of this doubly disadvantaged group. In contrast, high-educated women utilized primary – but not secondary - care to a greater degree than corresponding men, regardless of health care needs. Lastly, the moderately large utilization inequalities rooted in education were completely attributable to different health care needs, both for primary and secondary care. Taken together, the results paint a picture of primary and secondary care in Northern Sweden delivered according to needs when it comes to educational disparities, and with gender inequities disfavouring men remaining in primary care, but which appear equalized at entry to secondary health care.
These findings need to be viewed and interpreted in the light of the specific societal and health care context of Sweden, and can be expected to differ in other contexts depending on not only health care organization, financing and provision but also supporting welfare systems such as unemployment and parental leave benefits, the educational system and degree of gender equity in society.
Inconsistent impact of double disadvantage
One of the original tenets of intersectionality theory relates to the double jeopardy of multiple disadvantage - that “the intersectional experience is greater than the sum” (45). This notion has remained central in intersectionality-informed quantitative public health (25) and specifically operationalized in manners such as the excess intersectional disparity, originally defined by Jackson (38) as applied in this study. Whereas we found notable joint disparities observable throughout the analyses, they were not significantly different from the sum of the two referent disparities of gender and education, thereby not corroborating the double jeopardy hypothesis for these given outcomes and axes of inequality. It should be noted that the double jeopardy hypothesis indeed has been challenged as an oversimplified model, with conflicting empirical support (46) and critique for a simplified focus on “extreme groups” in any given intersectional space (35).
Nevertheless, our results unequivocally demonstrate that the doubly disadvantaged group of low educated women indeed report generally poor health and greater need of health care, which also corresponds to previous intersectionality-informed research on self-reported health from Sweden (32) and other European countries (47). However, beyond the pre-existing health disparities, we furthermore found that this manifest health disadvantage did not completely explain the high primary care usage in the doubly disadvantaged group (as seen in the joint inequity). For example, despite their quite distinct structural position, health profile, and lower crude utilization of health care, even higher primary care utilization given equal needs was reported by high-educated women (as seen in the referent gender inequity).
The central role of gender for primary health care utilization
As can be inferred from the discussion above, gender had a profound effect in shaping health care utilization in this study, particularly at the primary care level. This is consistent with other studies that reported higher utilization of health care amongst women as compared to men in high-income context (48-52). However, it contradicts findings from other intersectionality-informed research from the arguably quite different low-income context of India, where non-treatment of long-term ailments have been shown to be strongly patterned by gender across all economic classes but to the disadvantage of women instead of men (35).
The share of primary health care utilization not attributable to care needs among low- and high-educated women could possibly be explained by unobserved health care needs specifically relevant to women, such as maternity, gynaecological care and other aspects of women’s health. Previous Swedish studies have indeed found higher primary health care consumption among women than men, even when excluding health care for sex-specific morbidity and reproductive reasons for seeking care (8).However, the inequalities could also be explained by the impact of lower health care seeking behavior amongst men as compared to women (53), e.g. comparable to the previously reported difficulties to reach and engage Northern Swedish men for health promotion (54) or participation in patient education (55). In this sense, despite their socially advantaged position, men as a group are disadvantaged from seeking health care due to masculinity norms that may portray them as weak if they seek health care even if they are in need (48), also contingent on the intersection with e.g. socioeconomic position (53). On the other hand, one can also construe this observation as women using health seeking behaviors to successfully leverage the structural disadvantages of gender and low education, and resultantly partly compensate for their poor health.
Relative equity in specialist care utilization
The absence of horizontal inequities in specialist visits across all the four intersectional categories is in stark contrast to the substantial joint and referent gender inequities in general practitioner utilization. Due to the scarcity of intersectionality-informed studies it is difficult to make direct comparisons to the previous literature, and it is treacherous to make comparisons to studies focusing on only gender or educational inequities in health care utilization. However, one can note that this general inequity pattern contingent on level of care found in our study corresponds to previous Swedish studies reporting a considerably higher primary health care consumption among women than men, but only a marginal gender difference when it comes to specialist outpatient care (8), pro-rich horizontal socioeconomic inequities in primary but not secondary outpatient care (7), and higher utilization of general practitioners among low-income groups but higher utilization of private specialists among high-income groups (56). The results however contrast to other studies report inghigher educational inequities in outpatient specialist care than in general practitioner visits, e.g. from Norway (57) and across 12 European countries (57).
The comparatively equitable use of specialist visits in our study – both concerning gender, education, and their combination - could be reflective of the underlying forces that determine health care usage at each level of the Swedish health care system. Primary health care utilization is largely contingent on individuals’ own health-seeking behaviour, while access to specialist doctors is almost exclusively based on referrals from the primary health care level. This discrepancy when it comes to responsibility for reaching primary versus secondary care may be further compounded by reforms of primary health care over the last decade which place increased emphasis on the individual patient choices (3). As such, the access to primary health care is contingent on high- or low-educated women or men’s health literacy and differential health seeking behaviour, as discussed above (54, 55), but when inside the system, both men and women end up accessing specialists more equally because the decision lies with the primary health care doctors responsible for referrals. In this sense, our results could reflect an ‘equalizing’ effect of referral in the health care system in Northern Sweden that is linking those with greater health care needs at the primary level to specialist care.
On the other hand, it has been shown in multiple reports from Sweden that women are in fact disadvantaged when it comes to various specialist treatments, for example receive less expensive drug prescriptions, older dialysis treatment, later surgery for back pain, less services in case of Alzheimer’s disease, and overall lower quality of care (55). While these concerns specific health care outcomes not measured in this study, the gender ‘equalization’ apparent in our results at the level of specialist care could in fact reflect that women’s advantage at the primary care level is offset by the challenges faced in specialist care. Here it is also important to note that women’s relative higher health care consumption reflects low-cost care of primary care rather than the high-cost care of specialist care (8).
Education-related equity in health care utilization
The results showed no education-related intersectional inequities in accessing primary or secondary health care in Northern Sweden. This adds to previous studies on simple, non-intersectional, socioeconomic inequities in health care utilization from the same context, including small horizontal inequities in general practitioner visits, no inequities in specialist visit usage or hospitalizations (7), and among young adults, large income-related but no education-related inequities in youth clinics utilization (5, 6). A range of studies from other countries, as well as older studies from Sweden (58), have however reported higher health care utilization among high-educated or high-socioeconomic groups; e.g. across twelve European countries (59), and in several Low- or Middle-Income Countries such as Brazil (60), Mongolia (61), Nigeria (62), and Iran(63). While we indeed found large educational inequalities in both health and health care usage, they were in proportion to each other; i.e. health care utilization was commensurate to need, as posited by the principle of horizontal equity. The Swedish health care system is considered progressive and traditionally framed around the Beveridge model of health care financing, where health care is financed by general taxation thus promoting universal health access. Even though there has been a successively increased market-orientation and privatization of Swedish primary health care that may impact negatively on health care equity (3), Northern Sweden has been a region less affected by these developments (64). We conclude that health care at the primary health care level was utilized according to needs amongst intersectional groups of different educational level in this study. This finding could reflect the inherent impact of universal health coverage mitigating classism in the health system.
Methodological considerations
Although this study proposes a refinement to existing quantitative methods in assessing intersectionality in health care, we have noted some limitations that should be considered.
First, the response rate was 49%, which is comparable to most studies conducted in the same setting with reliable results. The demographic and social characteristics of the non-respondents (e.g. age, gender, education, area of residence) were not available and the extent of any selection bias could therefore not be assessed. Selection bias is therefore an uncontrolled threat against internal validity, and which could lead to either over- or underestimation of the studied inequities. Moreover, we cannot draw any causal inferences from our study as our data was collected from a cross-sectional survey
While the proposed analytical method produce an appropriate assessment of intersectional inequities on a familiar additive scale, which are not captured by conventional methods for horizontal inequities along single dimensions (41), it has limitations. Most importantly, it is developed and has only been used for two binary inequity dimensions (38), which limits the scope of the analysis. While it technically could be extended to include more inequity dimensions (e.g. ethnicity, sexual orientation, geographical, and disability), conceptualization, estimation and interpretation of the individual inequities becomes increasingly challenging. If the aim is to estimate a large set of inequity dimensions, other intersectionality-informed methods developed for this purpose might be more suitable (37). Moreover, estimation is more straightforward for continuous compared to binary outcomes (38), a limitation that however is even more pronounced for alternative methods (37). The method is furthermore based on adjustment for health care needs to assess horizontal inequity, and consequently, there is a risk of underestimation of health care needs as it is theoretically impossible to capture all health care needs. For instance, and as noted above, we could not provide adjustments for women’s health needs such as maternal health care needs, gynaecological requirements or other women reproductive health care needs, as this information was not available in the survey data. Nevertheless, we tried to capture several facets of health care needs that have also been applied in previous literature (5-7).